University of Massachusetts Amherst ScholarWorks@UMass Amherst Laying the groundwork for the DELV (Precursor NIH Working Groups on African American English literature, dissertations, joint work of the Working (AAE) Groups prior to the conceptualization of the DELV)

2004 Negative concord in child African American English: Implications for Specific Language Impairment D'Jaris Coles-White

Follow this and additional works at: https://scholarworks.umass.edu/aae_groundwork Negative Concord in Child African American English: Implications for Specific Language Impairment

D’Jaris Coles-White Wayne State University, In this study, African American English (AAE)-speaking children’s comprehension Detroit, MI of 2 different types of double negative sentences was examined and contrasted with that of a comparison group of Standard American English (SAE)-speaking children. The first type of double negative, negative concord, involves 2 negative elements in a sentence that are interpreted together as single negation. The second type of double negative, called true double negation, involves 2 negatives that are interpreted as independent negatives. A cross-sectional cohort of 61 (35 AAE, 26 SAE) typically developing children ranging in age from 5;2 (years;months) to 7;11 participated. The children responded to story- based grammatical judgment tasks that required them to differentiate between negative concord and true double negation. Results revealed no statistically significant differences between AAE- and SAE-speaking children in the way they interpreted negative concord and true double negation. However, there were significantly more correct responses to negative concord sentences across combined groups. In particular, the older children (i.e., 7-year-olds) produced more correct responses to negative concord than did the younger group (i.e., 5-year-olds). Explanations for these findings are framed in terms of children’s knowledge about sentences with 2 negatives, the constraints affecting the interpretation of 2 negatives that include negative concord, and the clinical importance of negative concord for assessing specific language impairment in child AAE speakers. KEY WORDS: negation, African American English, linguistic differences, language comprehension, negative concord

istorically, studies of African American English (AAE) have fo- cused on identifying and describing the features of the dialect. H Only recently have some studies shifted from description to ex- planation, using linguistic theory to account for specific feature func- tion and constraints underlying the grammar (Coles, 1998; Green, 1993; Jackson, 1998; Jackson et al., 1996; Johnson, 2001; Mufwene, Rickford, Bailey, & Baugh, 1998; Sells, Rickford, & Wasow, 1996). This shift to- ward examining the underlying mechanisms has been instrumental in moving the study of dialect forward to include studies that are clini- cally applicable for identifying language impairments (Craig & Wash- ington, 1994; McGregor, Williams, Hearst, & Johnson, 1997; Oetting & McDonald, 2001; Seymour, Bland-Stewart, & Green, 1998; Stockman, 1996; Washington, 1996). These recent investigations have addressed the challenges of differentiating typically developing AAE speakers from those use AAE and have specific language impairment (SLI), and

212 Journal of Speech, Language, and Hearing Research • Vol. 47 • 212–222 • February 2004 • ©American Speech-Language-Hearing Association 1092-4388/04/4701-0212

Downloaded from: https://pubs.asha.org University of Massachusetts on 41/27/2019, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions they have contributed to our knowledge base concern- concord, a -level syntactic construction with some ing the nature of AAE grammar. potential as a clinical marker of SLI in child AAE speak- In order to devise a truly comprehensive theory of ers. Clause-level syntactic processes such as negative SLI that takes into account AAE child speakers, the un- concord should be better linguistic markers for identi- derlying mechanisms of AAE and fying SLI in AAE-speaking children than morphologi- must be rigorously examined in both receptive and ex- cal markers, which are used for identifying SLI in Stan- pressive domains. A theory of SLI depends on develop- dard English dialects, because clause-level syntactic ing foundational knowledge about normal processes markers do not contrast with SAE. Accordingly, children used by child speakers and determining how these pro- may have an intuitive linguistic knowledge about nega- cesses are changed by children with SLI. For many Af- tive concord because it is a noncontrastive feature with rican American children this involves understanding SAE found in many early grammars (Bellugi, 1967; the processes underlying the use of major linguistic fea- Klima & Bellugi, 1973; Stokes, 1976), and by implica- tures of AAE, which may be contrastive or noncontras- tion its interpretation should be largely unaffected by tive with Standard American English (SAE) (Seymour AAE grammar. et al., 1998). Negative concord is a linguistic phenomenon that Contrastive features of AAE are those considered occurs in many standard and nonstandard dialects of to be different from the features that characterize SAE, languages. Negative concord is a complex, clause-level such as aspectual be (i.e., “Shaq be playin basketball”), syntactic process and is part of the negation system that where there is no interpretable complement or counter- many African American children develop in the acquisi- part in SAE. Those features that are noncontrastive have tion of AAE grammar. Negative concord may be defined similar or matching constituents with SAE, such as as the expression of two negative elements in a syntac- negative concord. For example, negative indefinite no, tic environment or sentence where they are in agree- when used in a negative sentence (i.e., “He don’t like no ment and, therefore, are interpreted together as a single sports”), does have an interpretable counterpart in SAE negation (Martin, 1992). The two negatives are not in- (i.e., “He doesn’t like any sports”). Seymour et al. (1998) terpreted separately as truly independent negatives (i.e., indicated that a focus solely on AAE features that are true double negation), but are dependent on each other contrastive with SAE may be problematic diagnostically, for a single negative reading. Example 1 illustrates nega- because on the surface contrastive features may appear tive concord and a true double negative sentence, where to be very similar to those features that characterize the negatives are interpreted independently. language impairment. However, a focus on noncontras- Example 1. tive features with SAE may protect against perceiving a. “He don’t have no friends” (“He doesn’t language difference (i.e., AAE) as language impairment have any friends” or “He has no friends”) and may help to establish what is vitally important for b. “He don’t like going there with no friends” characterizing language impairment, particularly with (“He doesn’t like going there without AAE-speaking children. friends”) Although some work has begun to contribute to our In 1a, both of the negatives are interpreted as a single understanding of the production of AAE features by chil- negation, thus a negative concord reading. However, in dren (e.g., Haynes & Moran, 1989; Hyter, 1996; Pollock 1b, the negative that occurs in the phrase “with no et al., 1998; Stockman & Vaughn-Cooke, 1992; Wash- friends” is interpreted independently of the negative ington & Craig, 1994) and the processes underlying these auxiliary don’t, which occurs earlier in the sentence and productions (Green, 1993), comprehension has been is thereby considered true double negation. largely ignored for its potential to inform us of what AAE speaking children know about language. Knowing that In developing grammars, negative concord is be- a child produces a sentence type does not necessarily lieved to share certain properties with interrogatives mean the child will fully comprehend the same sentence, under a concept of movement, which is essentially a particularly if it is heard without the context of familiar formal, linguistic explanation for how these two syn- routines and nonlinguistic cues (Miller & Paul, 1995, as tactic constructions are formulated in the deep struc- cited in Paul, 2001). ture (Haegeman, 1995; Martin, 1992; Progovac, 1994). Specifically, both constructions, negative concord and The purpose of this investigation was to contribute wh- questions, share or obey similar restrictions on syn- important, new information to this knowledge base by tactic contexts. These contextual restrictions have an examining comprehension of one major feature of child effect on the establishment of dependent relationships grammar, negation, by young typically developing AAE- between syntactic elements and ultimately influence speaking children. More specifically, this investigation the interpretation of negative concord sentences and examined and described the comprehension of negative

Coles-White: Negative Concord in Child African American English 213

Downloaded from: https://pubs.asha.org University of Massachusetts on 41/27/2019, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions wh- questions (see Chomsky, 1986, and Radford, 1988, 3. Were any of the items on the story-based grammati- for a review on syntactic movement operation; see Coles, cal judgment tasks more difficult for child AAE 1998, for a review on a concept of movement for nega- speakers to interpret than for the comparison group tive concord in AAE; see deVilliers & Roeper, 1995, for a of child SAE speakers? If so, do child AAE speakers review on barriers to wh- movement for young children). construe more of the minimally paired double nega- Studies have shown that responses to wh- questions tive sentences as negative concord? may have merit for assessing development of compre- hension skills in young AAE speakers (Craig & Wash- ington, 1994; Craig, Washington, & Thompson-Porter, Method 1998). Other studies have suggested that by examining Participants comprehension of wh- questions using grammatical judg- ment tasks, one may gain knowledge of children’s syn- Sixty-one typically developing children (29 boys and tactic understanding of structurally dependent relation- 32 girls) participated in the study. The target group con- ships and the constraints affecting interpretation of sisted of 35 African American, AAE-speaking children, complex constructions (Seymour, Bland, Champion, and the comparison group included 26 SAE-speaking deVilliers, & Roeper, 1992). If the formulation of nega- children. The comparison group had 25 European Ameri- tive concord in the deep structure is similar to that of can children and 1 biracial child of African American wh- questions and children’s knowledge of the structural and European American heritage. The children ranged properties of negative concord can be tested using gram- in age from 5;2 (years;months) to 7;11 (M = 6;4) and matical judgment tasks, then negative concord may also attended urban public schools in the northern United have merit for assessing the development of comprehen- States. Table 1 presents the participants’ descriptive sion skills in young AAE-speaking children. This seems information. particularly critical, as deficits in comprehension of com- plex syntax have been widely reported in children with Participant Selection language learning disorders (Gerber, 1993; Roth & Spekman, 1989). A 15–20 min sample of each child’s speech and lan- The following questions were addressed in this in- guage was videotaped prior to conducting the experi- vestigation: ment. The speech and language sample was elicited by means of a picture description task with 15 action pic- 1. Do child speakers of AAE perform differently than tures. Each participant was engaged by a bidialectal the comparison group of child speakers of SAE on African American doctoral student who was familiar story-based grammatical judgment tasks requiring with AAE and comfortable using some of the vocabu- them to understand the structural differences for lary and pronunciations styles of AAE when communi- minimally paired double negative sentences, where cating with African American participants. Specifically, one of the paired sentences should be interpreted as certain words were pronounced with intonation and negative concord and the other paired sentence stress patterns considered to be characteristic of many should receive a true double negation interpretation? AAE speakers and less characteristic of the majority of 2. Do child speakers of AAE and SAE vary in the fre- mainstream English speakers, such as changing the quency of correct responses to both types of double diphthong in “my” to a single vowel or monophthong, negative sentences? If so, do children differ in their “mah” (see Rickford & Rickford, 2000, for a review of ability to comprehend these constructions based on AAE vocabulary and pronunciation). The African Ameri- dialect, age, and gender? can doctoral student code-switched from SAE to AAE to

Table 1. Mean age in years and standard deviations of participants by age group, gender, and dialect.

AAE SAE

Male Female Male Female

Age group MSDMSDMSDMSD

5-year-olds 5.75 0.025 5.75 0.019 5.66 0.007 5.66 0.049 6-year-olds 6.16 0.010 6.33 0.044 6.25 0.031 6.33 0.028 7-year-olds 7.25 0.015 7.25 0.009 7.58 0.033 7.25 0.035

Note. AAE = African American English; SAE = Standard American English.

214 Journal of Speech, Language, and Hearing Research • Vol. 47 • 212–222 • February 2004

Downloaded from: https://pubs.asha.org University of Massachusetts on 41/27/2019, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions establish rapport when communicating with the Afri- set of double negative sentences and one single nega- can American participants and to indirectly acknowl- tive sentence. Each minimally paired set comprised one edge and permit the use of AAE. Many speakers of AAE true double negative (TN) sentence, where the negatives code-switch to SAE or a more standard variety of En- were not expected to be interpreted as a single nega- glish in situations where they perceive the use of AAE tion, and one negative concord (NC) sentence, where the to be unacceptable, such as with unfamiliar conversa- double negatives were expected to be interpreted in tional partners or listeners and in more formal settings, or as a single negation. Single negative sen- such as school (Rickford & Rickford, 2000). SAE was tences served as control items. The control item in each used when communicating with the European Ameri- story was consistent with that particular story line and can participants for similar reasons to establish rapport. corresponding pictures. Example 2, represented in Fig- Each child’s language sample was examined and ures 1–3, illustrates one of the stories with the mini- scored for the presence of grammatical and/or phono- mally paired set of double negative target sentences and logical features of AAE by graduate students trained to a single negative sentence. identify AAE dialect features. This brief interaction was Example 2. designed to establish the native dialect of each child. If a. A man wanted to feed a hungry baby. He any African American child used any of the phonologi- did feed a baby, but… He didn’t feed the cal or grammatical features consistent with AAE, he or baby with no hair. she was included in the target group. As a result, only b. He tried everything to feed the hungry those children whose ethnic backgrounds suggested that baby. He had a bottle and he had a spoon. AAE was their primary or native dialect—not those chil- He did feed the hungry baby, but… He dren who may use AAE features occasionally—were se- didn’t feed the baby with no bottle. lected. Similarly, only those children who used SAE as their native dialect and did not use any of the features c. The man wanted to feed the animals too, consistent with AAE were included in the comparison but he didn’t feed the one under the table group. Of the participants, all of the African American that meows. children used some phonological and/or grammatical fea- In 2a, the negatives are interpreted separately and a tures of AAE and were included in the target group. None true double negative response is expected, whereas in of the European American children used any AAE fea- 2b, the negatives are interpreted in agreement and a tures and were included in the comparison group. How- negative concord response is expected. In 2c, there is ever, there was one biracial child of African American and European American heritage who used only SAE and none Figure 1. A true double negative item from the DNc grammatical of the features of AAE when spoken to with AAE vocabu- judgment task. In the item depicted, a man is sitting at a table and lary and intonation patterns. Consequentially, the bira- preparing to feed one of two hungry babies; the baby with hair or the baby without hair. A correct response to the verbal prompt, cial child was included in the comparison group. “He didn’t feed the baby with no hair, which one did he feed?”, Group assignment was determined by primary dia- would be to point to the baby with hair. lect (i.e., AAE or SAE dialect). However, the children were also matched by age range and gender. The chil- dren were from working class backgrounds (based on residence, parental occupation, and/or level of educa- tion; principal/teacher/administrative official report; and participation in a reduced or free lunch program). School records, reports from the school speech-lan- guage pathologist, and teacher/administrator reports confirmed that all children selected for the target and comparison groups were native English speakers and free from any speech and/or language delay or disorder. None of the children selected had any hearing, cogni- tive, or social–emotional impairments.

Experimental Task The Double Negative Comprehension (DNc) Task featured five short stories presented with correspond- ing pictures. Each story comprised a minimally paired

Coles-White: Negative Concord in Child African American English 215

Downloaded from: https://pubs.asha.org University of Massachusetts on 41/27/2019, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions Figure 2. A negative concord item from the DNc grammatical room separate from his or her general classroom to avoid judgment task. In the item depicted, there are two feeding utensils, any noise or distractions and was then told, “I want you a spoon and a bottle, one of which must choose to feed a hungry to look at some pictures and listen to the stories that go baby. A correct response to the verbal prompt, “He didn’t feed the along with the pictures. Listen carefully, because there baby with no bottle, which one did he use?”, would be to point to will be some questions about each story that you will the spoon. have to answer.” Each child was presented with all five stories consecutively. All stories had three pictures pre- sented sequentially, matching the events in the stories. The stories were presented in random order. There was a fixed order of appearance for the minimally paired set of double negative target sentences because of story con- struction, which is illustrated above in Example 2. The control items were presented in random order to limit the effect of the fixed order of appearance for the mini- mally paired set of double negative target sentences. The children were asked to respond to questions about each story by pointing to some object in a picture. Adminis- tration of the entire DNc Task was approximately 20 to only one negative to interpret and a single negative read- 30 min for each child. ing is expected. All stories were prerecorded by the African Ameri- can experimenter so that there was no rise of intona- Reliability tion or stress placed on either of the negatives in the Responses were recorded online by the examiner, minimally paired set of double negative target sentences an African American doctoral student, and coded for and on the single negative in the control sentence to each target sentence as follows: 0 = marked interpre- influence children’s interpretation. tation of double negatives independently; 1 = marked interpretation of double negatives as negative concord. Twenty percent of each child’s responses were randomly Procedure selected for recoding in order to confirm the overall An African American doctoral student administered accuracy of coding decisions. The supervising doctoral the DNc Task to all children. Each child was taken to a student trained a master’s level graduate student in

Figure 3. A control item from the DNc grammatical judgment task. In the item depicted, a man is choosing to feed one of two animals, the cat or the dog, which are both beside the man’s feet. A correct response to the verbal prompt, “He didn’t feed the one that meows, which one did he feed?”, would be to point to the dog.

216 Journal of Speech, Language, and Hearing Research • Vol. 47 • 212–222 • February 2004

Downloaded from: https://pubs.asha.org University of Massachusetts on 41/27/2019, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions communication disorders at the University of Massa- Table 2. Mean scores and standard deviations on the double chusetts on coding responses prior to the recoding stage. negative comprehension task between children by dialect based on The graduate student reviewed participants’ responses correct responses to sentence types. from videotape to conduct reliability coding. Agreement was determined for scores of individual responses to Sentence types items. Interjudge agreement for all coding was very high, TN NC with over 90% for both double negative sentence types (i.e., 98% for TN items and 97% for NC items). The Dialect MSDMSD differences between the examiner and the reliability AAEa 3.51 1.07 3.97 1.01 coder were attributed to the differences in scoring re- SAEb 3.77 1.03 4.19 1.27 sponses online versus viewing a videotape. In the case Combined groupsc 3.62 1.05 4.07 1.12* of a disagreement, the examiner and the reliability coder reached an agreement by discussion. Note. Maximum score = 5.00 for each sentence type (e.g., TN and NC). AAE = African American English; SAE = Standard American English; TN = true negative sentences (negatives interpreted indepen- Data Analysis dently); NC = negative concord sentences (negatives interpreted as a single negation). Means and standard deviations were computed for an = 35. bn = 26. cn = 61. the correct responses on TN and NC sentence types of the DNc Task and for other grouping variables (i.e., dia- *There was a significant difference (i.e., p = .034) in performance on TN and NC sentence types for combined groups (i.e., AAE and SAE lect, gender, and age). To discover if the proportion of speakers). correct responses for each of the sentence types differed by dialect, paired t tests were run. To further examine group differences and determine if there were any in- planned comparison (Age 1) was defined as the differ- teractions among the groups for correct responses on ence between children 5 and 6 years old. The second the DNc Task, separate 2 × 2 × 3 (Dialect × Gender × comparison (Age 2) was defined as the difference be- Age) factorial analyses of variance (ANOVAs) were cal- tween children 5 and 7 years old. The third comparison culated for each of the sentence types. Next, in order to (Age 3) was defined as the difference between children find out whether specific items on the DNc Task were 6 and 7 years old. Results revealed that there were no more difficult for one dialect group than for the other, statistically significant interactions between any of the given variations in score distributions, a Mann–Whitney levels for TN sentence types. There was one significant U was computed. Finally, to assess whether groups dif- main effect for Age 2 contrasts with NC sentence types, fered on individual items on the DNc Task, logistic re- F(1, 49) = 4.70, p = .035. However, there were no other gression was calculated. significant main effects and no other interactions. The

Results Table 3. Mean scores and standard deviations on the double negative comprehension task between children by age based on Distribution Analysis correct responses to sentence types. There was no significance difference between dia- Sentence types lect groups (AAE and SAE) on the DNc Task. Mean scores for correct responses on the DNc Task revealed TN NC that children in this study understood the differences Age MSDMSD in meaning between sentences with two independent negatives (i.e., true double negation) and those sentences 5-year-oldsa 3.60 1.05 3.75 1.12 with two negatives in agreement (i.e., negative concord). 6-year-oldsb 3.63 1.01 3.92 1.21 A paired comparisons t test confirmed, however, that 7-year-oldsc 3.65 1.71 4.65 0.79* there were significantly more correct responses for the Note. Maximum score = 5.00 for each sentence type (e.g., TN and NC sentence types, t(60) = –2.17, p = .034, with a mod- NC). TN = true negative sentences (negatives interpreted indepen- erate effect size (d = .41). Table 2 presents the mean dently); NC = negative concord sentences (negatives interpreted as a scores and standard deviations for correct responses to single negation). TN and NC sentence types by dialect. an = 20. bn = 24. cn = 17. Accuracy by dialect, gender, and age for each sen- *There was a significant difference (i.e., p = .035) in performance on tence type (TN and NC) on the DNc Task was computed TN and NC sentence types for the older children (i.e., 7-year-olds) with separate 2 × 2 × 3 (Dialect × Gender × Age) facto- compared to the younger children (i.e., 5- and 6-year-olds). rial ANOVAs with planned comparisons on age. The first

Coles-White: Negative Concord in Child African American English 217

Downloaded from: https://pubs.asha.org University of Massachusetts on 41/27/2019, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions significant main effect for Age 2 contrasts with NC sen- tence types indicated that the older children, 7 years of Discussion age, had more correct responses. Table 3 presents the The outcome of this study demonstrated that typi- mean scores and standard deviations for correct re- cally developing children, both speakers of AAE and sponses on TN and NC sentence types by age. SAE, comprehend negative concord and that story-based grammatical judgment tasks are efficient at measuring young children’s knowledge of this complex linguistic Item Analysis feature. Children appear to be sensitive to the interpre- A nonparametric test was performed on the fre- tative differences necessary to distinguish negative con- quency of correct responses for the TN and NC sentence cord from sentences with two negatives as early as 5 types to determine if there were differences between years of age and, as expected developmentally, this abil- dialect speakers in the number of correct responses, tak- ity increases with age. ing into account differences in the score distributions. Overall, the children in this study performed simi- The Mann–Whitney U did not show any difference in larly on the DNc Task. This outcome suggests that AAE- rank scores by type of dialect spoken. Item difficulties speaking children generally understand the structural (e.g., proportion correct) for the TN sentence types differences for double negative sentences that are inter- ranged from .54 (SD = .50) to .93 (SD = .24) and from preted as negative concord and true independent nega- .77 (SD = .42) to .87 (SD = .34) for the NC sentence types. tion. However, the significant difference in overall cor- Logistic regression was used to examine differential item rect responses for NC versus TN sentence types may functioning (Swaminathan & Rogers, 1990). Each item indicate that negative concord is easier to interpret than was regressed on three variables: the total scale score true double negative sentences. Under binary choice (i.e., number correct for TN or number correct for NC), conditions with subtle differences between the struc- code for dialect group (i.e., AAE or SAE), and interac- tures, such as the experimental task used in this study, tion of Total Score × Dialect Group Code. No items were negative concord may be the default particularly when found to exhibit differential item functioning at p = .05. children are unsure of how to interpret double negatives. However, the odds ratio of 9.92 for NC Item 2 (p = .08) This may be because children have not yet learned the indicated that children in the SAE dialect group were complete negation system, which includes restrictions 9.9 times more likely to get that item number correct, on syntactic contexts for double negatives. Mastery of suggesting somewhat of a dialect effect for the one item basic negation and the rules of acquisition are well docu- on the DNc Task. Table 4 presents the means and stan- mented in the literature for young English-speaking dard deviations for the total score distributions and ad- children (e.g., Brown, 1973; deVilliers, 1979; Drozd, 1993; justed odds ratios for each item on the DNc Task. Klima & Bellugi, 1973). However, questions remain about the nature of syntactic negation and how children Table 4. Mean scores and standard deviations for the total score process superfluous negatives. It is not uncommon to distributions and the adjusted odds ratio (AOR) for items on the hear young children 4 and 5 years of age producing double negative comprehension task based on correct responses to doubled negative sentences (e.g., “I don’t want none”) sentence types. that are interpreted as negative concord regardless of dialect spoken. Therefore, in acquiring the rules of in- TN NC terpreting double negatives, negative concord is the first Item no. MSD AOR MSD AOR to be attained, followed by the rules for interpreting true double negation. The rules for interpreting negative con- 1 .79 .41 2.82 .79 .41 1.10 cord may be first because redundant information is 2 .54 .50 0.47 .84 .37 9.92a easier to process than additional negative information. 3 .70 .46 1.51 .80 .40 0.22 This notion of children acquiring different types of ne- 4 .66 .48 0.57 .77 .42 3.39 gation, including developing negative concord before 5 .93 .25 1.77 .87 .34 0.63 mastering true double negation, may be a universal pro- Note. TN = true negative sentences (negatives interpreted indepen- cess (Sharpe, Eakin, Cote, Lacroix, & Macnamara, 1996). dently); NC = negative concord sentences (negatives interpreted as a Similar processing patterns have been found in Polish- single negation). N = 61. Items that were less than one were easier for (Skowronski & Yan, 1992), Chinese- (Jou, 1988), and AAE-speaking children and items greater than one were easier for French-speaking children (Segui & Bertoncini, 1978, as SAE-speaking children. cited in Sharpe et al., 1996). aThe SAE-speakers were 9.9 times more likely to answer Item 2 The significant main effects for the Age 2 (i.e., 5- correctly than the AAE speakers, indicating a moderate language and 7-year-olds) contrast with NC sentence types showed effect (p = .08). that children in the older age group (i.e., 7-year-olds)

218 Journal of Speech, Language, and Hearing Research • Vol. 47 • 212–222 • February 2004

Downloaded from: https://pubs.asha.org University of Massachusetts on 41/27/2019, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions had more correct responses when compared to the Some of the AAE-speaking children may not have fully younger children. Although younger children interpreted understood the difference between the tools used for negative concord constructions, the older children were cutting and chopping, which may have prompted them better at judging when two negatives were to receive a to interpret the two negatives as a true double negative negative concord reading. Many features of grammar instead of the expected negative concord interpretation. are acquired by the time children are preschool age, 1 approximately 3 to 3 /2 years of age (e.g., Brown’s Stage III, Early Stage IV). However, double negatives, includ- Clinical Implications ing those interpreted as negative concord are a later developing grammatical feature, as are many aspects of The result of this investigation of children’s ability grammar, particularly syntax negation forms. Specifi- to comprehend negative concord has clinical implications cally, negative indefinites (e.g., nobody, no one, nothing), for the identification of SLI, specifically in AAE-speak- the determiner no used before nouns in a full sentence ing children. The current study addressed (a) syntactic (e.g., No cookies for you today), and negative derivational instead of the morphological aspects of AAE, (b) com- morphemes or such as un- and dis- are acquired prehension rather than expressive language skills, and after Brown’s Stage V; double negatives continue to be (c) one feature of AAE that is a noncontrastive feature difficult for children Stage V and beyond (Owens, 1996). of SAE. The late acquisition of double negation and the gradu- Children speaking AAE, unlike those speaking SAE, alness in further development once the concept begins exhibit specific characteristics of the grammar that con- to emerge may be due to the processing demands of in- tain unmarked or uninflected morphology. Some of these terpreting negative information, particularly double same characteristics are found in the language of chil- negation, versus interpreting affirmative information dren with SLI. Accordingly, using morphology to iden- (e.g., Sharpe et al., 1996). tify SLI in child AAE speakers presents a problem. Spe- The reliability of the grammatical judgment task in cifically, some of the morphological differences found in this study was acceptable when used with this sample the language of children with SLI may not be due to of AAE and SAE speaking children. Based on the total perceptual differences, missing features, deficit agree- score distribution, no one item was any more difficult ment systems, or an extended optional infinitives pe- than any other for the children to understand. Similarly, riod. Rather, morphological differences may be due to there was no statistical significance in the degree of dif- an inability to create syntactically complex utterances. ference for items on the grammatical judgment task. That is, some of the morphological differences found in Both AAE and SAE speakers performed similarly on the language of children with SLI may be due to an un- items. However, the odds ratio of 9.92 for NC Item 2 (p derlying deficit involving certain syntactic operations. = .08) indicated that children in the SAE dialect group Therefore, syntactic explanations for the morphological were 9.9 times more likely to get that item number cor- differences found in the dialect of children speaking AAE rect, suggesting somewhat of a dialect effect for the one may provide additional diagnostic support for identify- item on the DNc Task. This item was constructed simi- ing SLI because studies have found that children with larly to the other items across the DNc Task. Despite SLI have difficulty with syntax. that, NC Item 2 may have contained a semantic ambi- A study done by van der Lely (1998) suggested that guity. The choice of words for this story or the phrasing the linguistic characteristics found in the language of of the minimally paired double negative sentences may children with grammatical SLI might reflect problems have contributed to this near-significant difference be- with the syntactic computational system, which includes tween AAE- and SAE-speaking children. Example 3 il- problems with the operations involving syntactic move- lustrates the story and the minimally paired double ment. Accordingly, some children with SLI may exhibit negative sentences for Item 2 and the control item. difficulties with syntactic movement operations, which Example 3. notably reduce the complexity of grammar. Specifically, structurally complex grammar includes wh- questions, A boy wanted to cut down something with and passives, which necessitate syntactic movement op- his tools. He did cut something, but… He erations. Deficits underlying such grammatical struc- didn’t cut down the fence with no gate. He tures contribute to the general syntactic simplicity of the tried everything to cut down the fence. He utterances of children with SLI. The results of the cur- had an axe and he had a saw. He did cut rent study suggest that negative concord is represented down the fence, but… He didn’t cut down the and interpreted in the grammar similarly to wh- ques- fence with no axe. He wanted to cut down a tions, which involve similar movement operations. There- tree, but… He didn’t cut down the tree with fore, if a child has difficulty in comprehending questions, leaves. that child may also have difficulty comprehending

Coles-White: Negative Concord in Child African American English 219

Downloaded from: https://pubs.asha.org University of Massachusetts on 41/27/2019, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions negative concord. Connecting negative concord with More participants would have provided more inferen- other syntactic operations involving movement may have tial power for interpreting performance between groups. implications for identifying SLI particularly in child Second, more items and better illustrated stories might AAE speakers. have resulted in more precise responses for the sentence Comprehension data may provide a more effective types and might have provided more information regard- way of investigating children’s difficulties with language, ing item functioning across groups. Better stories for especially data from child AAE speakers compared to both sentence types might have eliminated some of the production data. There are a number of syntactic rules difficulty or degree of difficulty with interpreting true and regulations that characterize AAE grammar and double negatives. account for the great variability in the use of AAE (Craig Further exploration of negative concord is needed & Washington, 1994). Examination of the comprehen- with a larger sample of typically developing, as well as sion of AAE rules in a decontextualized format, such as with language impaired, child AAE speakers. Future with grammatical judgment tasks, should result in di- directions include extending this study to include in- minished variability of responses and present a more vestigation of other negative morphology that would transparent observation of the child’s underlying pro- entail negative concord (e.g., negative indefinites, such cesses or knowledge of AAE grammar. This may be par- as no one, nobody, nothing) and true double negation ticularly true of complex syntax constructions involving (e.g., derivational morphemes or negative prefixes, such movement operations. Craig and Washington (1994) and as un- and dis-) to further document specific feature Craig et al. (1998) revealed that child AAE speakers with functioning and the possibility of double negative sen- SLI had difficulty understanding complex sentences. By tence types, specifically negative concord as a clinical implication, examining production of negative concord marker for identifying SLI in child AAE speakers. may not show children’s knowledge of the underlying processes that affect interpretation of this complex clause-level process as well as grammatical judgment Acknowledgments tasks, because production data can vary from child to The research in this report was supported by National child and from dialect to dialect. Therefore, by testing Institute on Deafness and Other Communication Disorders comprehension skills the type and the variety of utter- (NIDCD)/Office of Research on Minority Health Grant ances found in children’s expressive language might be R03DC97002 as part of the fulfillment for the degree of controlled for more easily. Also, a feature such as nega- Doctor of Philosophy at the University of Massachusetts at Amherst in the Department of Communication Disorders. tive concord may not be present consistently and regu- Research report preparation was supported in part by larly in a general conversational exchange or narrative, NIDCD Grant K23 DC 00181. but a grammatical judgment task testing knowledge of Acknowledgments are extended to Harry Seymour, Tom negative concord may more readily provide enough data Roeper, and Jill de Villiers. I am grateful for input and to draw some important conclusions about children’s lan- discussion provided by them on an earlier version of this guage regardless of the dialect spoken. research report submitted as a doctorial dissertation. Lastly, negative concord is one feature of AAE that Acknowledgments are also extended to Julie Washington may be considered a noncontrastive feature with SAE and Ljiljana Provogac for their helpful comments and mentoring while I was working on the current version of this because the properties of negative morpheme no appear research report. Further appreciation is extended to the to function similarly to indefinite morpheme any under principals and teachers of elementary schools, and the operations of negation. Consequently, the underlying families of the participants in Springfield, MA, and Hart- constraints or conditions affecting the interpretation of ford, CT, for their support of this project. no seem to apply equally for any under operations nega- tion. The results of this study revealed that negative concord is interpretable by both AAE- and SAE-speak- References ing children. Thus, syntactic operations, rather than Bellugi, U. (1967). The acquisition of negation. Unpublished morphology, may be a more appropriate way of address- doctoral dissertation, Graduate School of Education, ing those features of a child’s grammar that are noncon- Harvard University, Cambridge, MA. trastive and consistent across dialects. Brown, R. (1973). A first language: The early stages. Cambridge, MA: Harvard University Press. There are some limitations of this study that war- rant discussion. These limitations include a small num- Chomsky, N. (1986). Barriers. Cambridge, MA: MIT Press. ber of participants and a limited number of task items Coles, D. (1998). Barrier constraints on negative concord in with some illustrative difficulty. Although there were African American English. Unpublished doctoral disserta- tion, University of Massachusetts, Amherst. 61 participants in this study, a larger data set would have made a more representative population sample. Craig, H., & Washington, J. (1994). The complex syntax skills of poor, urban, African-American preschoolers at

220 Journal of Speech, Language, and Hearing Research • Vol. 47 • 212–222 • February 2004

Downloaded from: https://pubs.asha.org University of Massachusetts on 41/27/2019, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions school entry. Language, Speech, and Hearing Services in (Eds.). (1998). African-American English: Structure, Schools, 25, 181–190. history, and use. New York: Routledge. Craig, H., Washington, J., & Thompson-Porter, C. Oetting, J., & McDonald, J. L. (2001). Nonmainstream (1998). Performances of young African-American children dialect use and specific language impairment. Journal of on two comprehension tasks. Journal of Speech, Lan- Speech, Language, and Hearing Research, 44, 207–223. guage, and Hearing Research, 41, 445–457. Owens, R. (1996). Language development: An introduction deVilliers, J. G. (1979). Form and function in the develop- (4th ed.). Boston: Allyn and Bacon. ment of sentence negation. Papers and Reports on Child Paul, R. (2001). Language disorders from infancy through Language Development, 17, 57–64. adolescence: Assessment and intervention. St. Louis, MO: deVilliers, J. G., & Roeper, T. (1995). Relative are Mosby. barriers to wh-movement for young children. Journal of Pollock, K., Bailey, G., Berni, M., Fletcher, D., Hinton, Child Language, 22, 389–404. L., Johnson, I., & Weaver, R. (1998, November). Drozd, K. F. (1993). A unification categorical grammar of Phonological characteristics of African American Vernacu- child English negation. Unpublished doctoral dissertation, lar English (AAVE): An updated feature list. Seminar University of Arizona, Tucson. presented at the convention of American Speech-Language- Gerber, S. (1993). Language-related learning disabilities: Hearing Association, San Antonio, TX. Their nature and treatment. Baltimore: Paul H. Brookes. Progovac, L. (1994). Negation and positive polarity: A Green, L. (1993). Topics in African American English: The binding approach. Cambridge, MA: Cambridge University verb system analysis. Unpublished doctoral dissertation, Press. University of Massachusetts, Amherst. Radford, A. (1988). Transformational grammar: A first Haegeman, L. (1995). The syntax of negation. Cambridge, course. Cambridge, MA: Cambridge University Press. MA: Cambridge University Press. Rickford, J. R., & Rickford, R. J. (2000). Spoken soul: The Haynes, W., & Moran, M. (1989). A cross-sectional develop- story of Black English. New York: Wiley. mental study of final consonant production in southern Roth, F., & Spekman, N. (1989). Higher-order language Black children from preschool through first grade. processes and reading disabilities. In A. Kamhi & H. Catts Language, Speech, and Hearing Services in the Schools, (Eds.), Reading disabilities: A developmental language 20, 400–406. perspective (pp. 159–198). Boston: College-Hill. Hyter, Y. (1996). Ties that bind: The sounds of African Sells, P., Rickford, D., & Wasow, T. (1996). Negative American English. American Speech-Language-Hearing concord in African American vernacular English. Natural Association Special Interest Division 14 Newsletter, 2, Language and Linguistic Theory, 14, 591–627. 3–6. Seymour, H., Bland, L., Champion, T., deVilliers, J. G., Jackson, J. E. (1998). Linguistic aspect in African-Ameri- & Roeper, T. (1992, November). Wh- movement in can English speaking children: An investigation of children of divergent language backgrounds. Seminar aspectual be. Unpublished doctoral dissertation, Univer- presented at the annual convention of American Speech- sity of Massachusetts, Amherst. Language-Hearing Association, San Antonio, TX. Jackson, J., Ramos, E., Hall, F., Coles, D., Seymour, H., Seymour, H., Bland-Stewart, L., & Green, L. (1998). Dickey, M., et al. (1996). They be taggin’, don’t they? The Difference versus deficit in child African American acquisition of invariant be. In A. Stringfellow, D. Cahana- English. Language, Speech, and Hearing Services in Amitay, E. Hughes, & A. Zukowski (Eds.), BUCLD 20 Schools, 29, 96–108. Proceedings 1 (pp. 364–373). Somerville, MA: Casacadila Sharpe, D., Eakin, L., Cote, M., Lacroix, G., & Press. Macnamara, J. (1996). The acquisition of negation Johnson, V. E. (2001). Fast mapping verb meaning from interpretation rules. In A. Stringfellow, D. Cahana- argument structure. Unpublished doctoral dissertation, Amitay, E. Hughes, & A. Zukowski (Eds.), BUCLD 20 University of Massachusetts, Amherst. Proceedings 2 (pp. 694–705). Somerville, MA: Casacadila Jou, J. (1988). The development of comprehension of double Press. negation in Chinese children. Journal of Experimental Skowronski, T., & Yan, G. (1992). Cross-cultural test of an Child Psychology, 45, 457–471. asymmetry of language knowledge dimensions: A Polish- Klima, E. S., & Bellugi, U. (1973). Syntactic regularities in Chinese study. Polish Psychological Bulletin, 23, 263–281. the speech of children. In C. A. Ferguson & D. I. Slobin Stockman, I. (1996). The promises and pitfalls of language (Eds.), Psycholinguistics papers (pp. 333–354). Edinburgh, sample analysis as an assessment tool for linguistic U.K.: Edinburgh University Press. minority children. Language, Speech, and Hearing Martin, S. (1992). Topics in the syntax of nonstandard Services in Schools, 27, 355–366. English. Unpublished doctoral dissertation, University of Stockman, I., & Vaughn-Cooke, F. (1992). Lexical Maryland, College Park. elaboration in children’s locative action constructions. McGregor, K., Williams, D., Hearst, S., & Johnson, A. Child Development, 63, 1104–1125. (1997). The use of contrastive analysis in distinguishing Stokes, N. H. (1976). A cross sectional study of the acquisi- difference from disorder: A tutorial. American Journal of tion of negation structures in Black children. Unpub- Speech-Language Pathology, 6(3), 45–56. lished doctoral dissertation, Georgetown University, Mufwene, S., Rickford, J., Bailey, G., & Baugh, J. Washington, DC.

Coles-White: Negative Concord in Child African American English 221

Downloaded from: https://pubs.asha.org University of Massachusetts on 41/27/2019, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions Swaminathan, H., & Rogers, H. J. (1990). Detecting Washington, J., & Craig, H. (1994). Dialectal forms during differential item functioning using logistic regression discourse of poor, urban, African American preschoolers. procedures. Journal of Educational Measurement, 27, Journal of Speech and Hearing Research, 37, 816–823. 361–370. van der Lely, H. K. J. (1998). SLI in children: Movement, Received February 19, 2003 economy, and deficits in the computational-syntactic system. Language acquisition, 72(2–4), 161–192. Accepted June 16, 2003 Washington, J. (1996). Issues in assessing the language DOI: 10.1044/1092-4388(2004/018) abilities of African American children. In A. Kamhi, K. Contact author: D’Jaris Coles-White, PhD, Department of Pollock, & J. Harris (Eds.), Communication development Audiology and Speech-Language Pathology, Wayne State and disorders in African American children (pp. 35–54). University, 906 West Warren Avenue, 581 Alex Manoogian Baltimore: Paul H. Brookes. Hall, Detroit, MI 48202. E-mail: [email protected]

222 Journal of Speech, Language, and Hearing Research • Vol. 47 • 212–222 • February 2004

Downloaded from: https://pubs.asha.org University of Massachusetts on 41/27/2019, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions